On 2021-11-6 10:14 , Nils Breunese wrote:
Joshua Root <[email protected]> wrote:

On 2021-11-5 18:42 , Ryan Schmidt wrote:
Either we allow portindexes that differ by OS version and arch (and we 
currently do), or we do not.

That's something of a false dichotomy; we can also allow it for some but not all options. 
I am advocating disallowing it by policy for version and revision, not just because of 
this issue but because it makes it impossible to answer questions like "What is the 
current version of this port?"

I maintain the various openjdk* ports and the latest versions for x86_64 and 
arm64 are not always the same for a particular openjdk* port. So what the 
current version of a port is cannot always be answered without the context of 
OS/arch.

Yes, but that need not be the case. Earlier in the thread I mentioned the approach used by ld64, where each version is a separate subport, and the top-level port depends on a different subport depending on the OS version.

- Josh

Reply via email to